PicoScope 7 Software
Available on Windows, Mac and Linux
triple-value reading
triple-value reading
I have several instruments of usb TC-08. I noticed a common problem in some of them: the derivative of the temperature with time has a noise with a frequency of ~0.11Hz. the amplitude and the exact frequency of the noise change over time. see attached figure.
I see the same phenomena on several pico units used in several locations.
Any suggestion on how to fix\prevent this noise?
I see the same phenomena on several pico units used in several locations.
Any suggestion on how to fix\prevent this noise?
- Attachments
-
Re: triple-value reading
Hi LBN,
I'm responding to this quite late, but if this was not resolved for you, then I this answer may be helpful.
You haven't provided any information regarding the sample rate that you are capturing the data at. Your plotted derivative looks like it has some kind of waveform envelope, which is what you can get with a sample rate that is too low. I've had some issues with creating a derivative plot in PicoLog 6, but I may be able to help by just looking at the data you have.
For the purpose of discussion, let's assume that you're using a sample rate of 100ms, with 8 Thermocouples per TC-08. You will get approximately 9 total conversions that are required, per second (because the Analog-to-Digital Converter (ADC) is shared among channels, and each Thermocouple conversion time is 100ms, but you also need to convert the Cold Junction temperature, which is an extra 100ms). So, for any one thermocouple, the fastest possible conversion time will be 900ms (the time taken to for the ADC to convert all of the values before getting back to the 1st Thermocouple). So, any one Thermocouple is sampled at a rate of 1.11S/s (which happens to be 10 times the frequency that you mentioned).
Looking at your derivative graph it looks like the maximum amplitude of the anomaly (that you call noise, but which is more likely a function of the sampling process, as noise doesn't have one frequency) is at the point where the rate of change of the data is the largest and the minimum amplitude is where the rate of change is the smallest, which also suggests that the anomaly is a function of the sampling process (because the bandwidth of the data is highest at that point, so erroneous plotting of differentiated data would have the biggest affect there).
There's not much point in continuing the discussion as I don't know exactly what your set-up, or what your sample rate is, and don't know what the original captured data looked like (you could post these if you want us to look into this further). However, this gives you some idea of where to look to try correcting the anomaly (i.e. you may need to just increase the number of points used for your plotted Derivative graph, if you are using a tool to plot the Derivative of your captured data at a rate that is different to the sample rate used to capture the data from any one Thermocouple.
Regards,
Gerry
I'm responding to this quite late, but if this was not resolved for you, then I this answer may be helpful.
You haven't provided any information regarding the sample rate that you are capturing the data at. Your plotted derivative looks like it has some kind of waveform envelope, which is what you can get with a sample rate that is too low. I've had some issues with creating a derivative plot in PicoLog 6, but I may be able to help by just looking at the data you have.
For the purpose of discussion, let's assume that you're using a sample rate of 100ms, with 8 Thermocouples per TC-08. You will get approximately 9 total conversions that are required, per second (because the Analog-to-Digital Converter (ADC) is shared among channels, and each Thermocouple conversion time is 100ms, but you also need to convert the Cold Junction temperature, which is an extra 100ms). So, for any one thermocouple, the fastest possible conversion time will be 900ms (the time taken to for the ADC to convert all of the values before getting back to the 1st Thermocouple). So, any one Thermocouple is sampled at a rate of 1.11S/s (which happens to be 10 times the frequency that you mentioned).
Looking at your derivative graph it looks like the maximum amplitude of the anomaly (that you call noise, but which is more likely a function of the sampling process, as noise doesn't have one frequency) is at the point where the rate of change of the data is the largest and the minimum amplitude is where the rate of change is the smallest, which also suggests that the anomaly is a function of the sampling process (because the bandwidth of the data is highest at that point, so erroneous plotting of differentiated data would have the biggest affect there).
There's not much point in continuing the discussion as I don't know exactly what your set-up, or what your sample rate is, and don't know what the original captured data looked like (you could post these if you want us to look into this further). However, this gives you some idea of where to look to try correcting the anomaly (i.e. you may need to just increase the number of points used for your plotted Derivative graph, if you are using a tool to plot the Derivative of your captured data at a rate that is different to the sample rate used to capture the data from any one Thermocouple.
Regards,
Gerry
Gerry
Technical Specialist
Technical Specialist